FORUM on Corrections Research

Exploring reintegration potential: Impacts of initial placement practice

Canadian incarceration rates remain among the highest reported by western or industrial countries^{(2 )}despite: decreasing rates of crime⁽³⁾ and admissions to federal custody;⁽⁴⁾ new legislation to expedite the release of low-risk offenders (Bill C-55); modern risk assessment technology; new programs⁽⁵⁾ to improve the release potential of offenders; and a growing acceptance of alternatives to incarceration. Although the institutional and conditional release populations of the Correctional Service of Canada have decreased recently, the decrease is largely attributable to a sharp decline in Warrant of Committal admissions, a by-product of the falling crime rate. In response, the Service has adopted a reintegration strategy to increase discretionary releases of appropriate offenders that includes: reconfiguring organizational structures and responsibilities; emphasizing "reintegration potential" rather than "risk assessment"; eliminating duplication of effort to streamline case management functions; applying institutional programs judiciously so that they target only criminogenic factors; giving case management officers broader discretion and more latitude to use professional judgment; and encouraging efficiency throughout the case management process.

Improving decision making and efficiencies

The centrepieces of the reintegration strategy can be boiled down to two key tactics -- identifying release potential and expediting release. With the introduction of accelerated full and day parole reviews, many of the most obvious candidates for reintegration are now being released by statute, leaving the difficult task of identifying and developing reintegration potential in a more tenuous candidate pool. The Service relies on a variety of system-wide, objective measures to assess security classification, risk and needs, psychological profiles, and recidivism potential, and these measures are central to the identification of reintegration potential. Objective measures offer a method for finding more offenders suitable for release, relating reintegration potential to standardized risk profiles, and monitoring, assessing and refining decision-making protocols. For example, a recent survey of Offender Intake Assessment Profiles on the Offender Management System (OMS)⁽⁶⁾ revealed that, of the many offenders past their parole eligibility date, 319 were rated "low-risk/low-need," 216 were rated "low-risk/medium-need," and 405 were rated "medium-risk/low-need" by their current case management officer. On average, the low-risk/low-need group were 307 days past parole eligibility, the low-risk/medium-need group were 261 days past, and the medium-risk/low-need group were 131 days past. Releasing on the parole eligibility date and improving the rate of discretionary releases (for example, by 50% for the low-risk/low-need group; 30% for low-risk/medium-need group; and 20% for the medium-risk/low-need group), could increase the total of days on conditional release by 80,000 days, and add 210 full-year cases to community caseloads.

Similar surveys indicate variations across the Service's five regions in institutional-community population distributions, in suspension and revocation rates, and in days under temporary detention. These regional and responsibility centre disparities in case management functioning suggest that contributions to reintegration could be realized by improving efficiencies and encouraging practices used at the most efficient units. The accumulated impact of a general efficiency model, designed to achieve even modest improvements throughout all case management functions, and targeting offenders for release by indexed assessment, can lead gradually to substantial increases in community caseloads.

To illustrate the potential impact on reintegration of appropriate, efficient decision making, a detailed analysis of initial placement practice was undertaken using Custody Rating Scale (CRS) data. Initial placement practice is particularly useful for investigation because: it is governed by a specific policy to pursue the least restrictive measures of confinement; it employs an objective method of initial classification; and it uses electronically accessible data that are complete and reliable enough to permit relatively competent projections and assertions. In the rest of this discussion, the contribution of modest changes to initial placement practice to reintegration is illustrated.

Initial placement and reintegration

In the past 25 years, objective methods of initial security classification and reclassification have proliferated throughout correctional jurisdictions.⁽⁷⁾ When compared with traditional classification procedures, objective measures have been credited with lowering security classification profiles without adverse effect on the rates of prison misconduct, escapes and fatalities.^{(8 )}In a recent Correctional Service of Canada study⁽⁹⁾ comparing initial placement practice with CRS scores, similar findings were reported. The study found that the scale assigned a higher proportion of offenders to lower security categories than actual initial placement, and was more accurate in predicting incidence of institutional maladjustment and escape from minimum security. Of particular importance to the reintegration strategy was the finding that offenders with similar risks, as measured by both the CRS and the Statistical Information on Recidivism (SIR) Scale, but initially placed in different security levels, were granted discretionary release at substantially different rates and varied widely in the number of days served before release. For example, the discretionary release rates of offenders rated and placed in minimum security was 17% higher, and they received their release an average of 77 days earlier, than offenders with similar CRS and SIR ratings placed in medium security. (Conversely, the release rate of offenders rated medium security but initially placed in minimum security was higher than that of the minimum-rated but medium-placed group, despite the lower risk ratings of the latter.) The results demonstrate that the security level of an offender's initial placement significantly affects both the likelihood and the timing of discretionary release. More specifically, the lower the level of initial security placement, the greater the probability of discretionary release and the shorter the period of incarceration before release. It clearly suggests the reintegration potential of initial placement.

The relationship between discretionary release, earlier release and initial placement to lower security level, particularly minimum-security placement, is not difficult to explain. Initial placement to any of the Service's unfenced minimum-security facilities places a burden of trust on the offender not unlike that involved in conditional release. Successful adjustment to minimum security contributes substantially to the offender's release credibility. This fact is formally recognized by the SIR scale, which assigns higher risk ratings to inmates of maximum-security institutions. The sharp increase in minimum-security program opportunities since 1992 has allowed many more offenders to be initially placed in minimum-security institutions and receive a better preparation for release. Initial placement at the lowest security levels avoids the time-consuming and often unnecessary reclassification and institutional transfer process -- the route that many successfully released candidates traditionally take to minimum-security institutions. Finally, case management teams at minimum-security institutions tend to be much more release-oriented, usually because they have less time to process large numbers of higher-quality release candidates and, as a result, the case preparation regimen is expeditious and highly efficient.

The impact of modest changes to initial placement

In April 1997, CRS, placement and release data were drawn from OMS on all offenders for whom a CRS was prepared in the six years (1991 to 1997) that the CRS was in use. The majority of these data represent the more recent years, when use of the CRS was promoted. The data were used to recalculate the discretionary-release and days-before-release rates to determine the differential effects of various CRS and placement options. Table 1 shows the concordance rates between CRS security level ratings and actual penitentiary placements for more than 14,000 files.

Table 1

Penitentiary Placement by CRS Rating
		Security Level of Placement			CRS Distribution
		Maximum	Medium	Minimum
Maximum CRS Rating	Frequency	451	404	10	865
	% of row	52.1	46.7	1.2
	% of total n	3.2	2.9	0.07	6.15
Medium CRS Rating	Frequency	562	7,777	1,115	9,454
	% of row	5.9	82.3	11.8
	% of total n	4.0	55.3	7.9	67.3
Minimum CRS Rating	Frequency	33	1,613	2,090	3,736
	% of row	0.9	43.2	55.9
	% of total n	0.2	11.5	14.9	26.6
Placement Distribution	Total n	1,046	9,794	3,215	4,055
	% of total	7.4	69.7	22.9	100

The overall CRS-placement concordance rate, as represented by the sum of the total percentages on the diagonal, is 73%, a reassuring rate of agreement, given that 8% to 12%⁽¹⁰⁾ of the disagreements may arise from non-security considerations such as a need for protection, medical treatment, programs or availability of cell accommodation at the rated level. Focusing specifically on minimum security, the CRS rated a higher proportion of offenders as minimum security (26.6%) than were actually placed there (22.9%). Overall concordance is a reflection of the substantial agreement between medium-security-rated and medium-security-placed offenders (82.3%), but the most striking finding, particularly from the perspective of reintegration, was the large proportion of offenders rated minimum security who were nevertheless placed in medium-security institutions (43.2% or 1,613). While this group is partly offset by offenders rated medium security who were placed in minimum security (1,115), such placements tend to produce increases in escape and institutional incident rates.⁽¹¹⁾

Next we turn to the potential impact on discretionary release of improving concordance with the CRS scores of offenders rated minimum security. In Table 2, we recalculated the discretionary release rates and average days before release for offenders rated minimum security by the CRS, based on data drawn in April 1997.

The discretionary release rate of offenders rated minimum security and placed in minimum security is 84% and the average number of days before release was 311. This compares with a 62% release rate and 376 days before release for offenders rated minimum security but placed in medium security. The results confirm those reported earlier and, again, demonstrate that offenders rated but not placed in minimum can experience as much as a 22% differential in discretionary release potential and, on average, remain incarcerated 65 days longer.

As noted in Table 1, as many as 1,613 (43%) of offenders rated minimum security by the CRS were placed in medium security. If we assume that between 8% and 12% of these placements result from non-security risk factors rather than a basic disagreement with the CRS rating, this leaves many minimum-rated offenders placed in medium security who could have been placed according to their CRS rating.

What is the potential for initial placements in minimum security? Table 3, the differential release and average days-before-release rates are applied to the annual average CRS override and admission rates to illustrate the regional and national impact of improving concordance to 80% of offenders rated minimum security. The impacts are a function of the number of projected admissions and the current rate of overrides of offenders rated minimum security. As the table shows, the impacts vary sharply across the regions. By way of explanation, the Atlantic region averaged annually approximately 1,054 offender admissions from 1993 to 1997, of which 9.56% were rated minimum security but were placed in medium security. By improving the Atlantic region's concordance rate with offenders rated minimum security to 80%, 81 more offenders may have been initially placed in minimum security. Thus, if the rest of their sentence went well, an additional 22% or 18 offenders, would be released if the full placement effect is realized. Finally, if all additional offenders placed in minimum security were released earlier, a total of 4,423 days of incarceration would be saved.

Table 2

Effect of Security Level of Placement on length of incarceration
	Minimum Rated Minimum Placed	Minimum Rated Medium Placed	Estimated Gain in Release Rate	Estimated Reduction in Days of Incarceration per Additional Release
Discretionary Release Rate	84%	62%	22%	-
Days of incarceration	311	376	0,07	65

Additional adjustments can also be made to the security level cutoff values of the CRS to allow for an increase in the distribution to minimum security or to customize distributions to fit the unique needs of individual regions. In the current study, approximately 26% of the national sample was assigned to the minimum-security category but, by raising the institutional adjustment sub-scale by 2.8 points (from 79.5 to 82.2 points) and the security risk sub-scale by 2 points (from 58.5 to 60.5 points), an additional 4% or 30% of offenders, would be minimum-security candidates. By adjusting the CRS to allow for a 30% minimum-security distribution and using an 80% concordance rate, the minimum-security candidate pool increased by 1,066 offenders, with concomitant potential to reduce days before release by more than 57,000 and increase discretionary releases by 234 offenders. The risk associated with the new cutoff values is minor relative to the original values, and the actual performance of the more liberal cutoff values can be carefully tracked. (In a simulation exercise using the sample developed for the CRS validation study cutoff values, similar to those identified above, which allowed for a 30% minimum-security distribution, resulted in no change in the escape rate of offenders initially placed in minimum security over the escape rate reported for the original cutoff values.)

Table 3

Net Annual Gain in Minimum-security Population, Discretionary Release and Reduction in Days of Incarceration at 80% CRS Concordance
	Atlantic	Quebec	Ontario	Prairie	Pacific	National
Average annual admission (1993-1997)	1 054	2 443	2 041	2 218	979
Minimum to medium override rate 1995-1996 admissions*	9.56%	13.26%	9.43%	16.93%	13.71%
Net gain in minimum population with 80% concordance**	81	259	154	300	107	901
Net reduction in days of incarceration***	4 423	14 141	8 408	16 380	5 842	49 194
Net gain in discretionary release****	18	57	34	66	24	199
Notes: national totals are sum of regioanl numbers * Proportion of total annual admissions rated minimum but overriden to medium -security **Projections based on a reduction in regional overrides to 20% of rates reported in 1995-1996 ***Projection based on 65 days reduction per additional minimum-security placement resulting from improving concordance with minimum-rated offenders to 80% ****Projections based on a 22% increase in discretionary release of appropriate offenders placed in minimum security

Conclusions

Clearly, the number of additional placements and releases and the reduction in incarceration days cannot be precisely calculated until the number of non-security overrides is known, the effects of the placement practices and traditions of each region are accounted for, and adjustments are made to factor in implementation effects and accommodation restrictions. Despite the impediments to realizing the full potential for releases, efforts to encourage initial placement in minimum security not only advance the principle of "least restrictive measure of confinement," but are also good correctional practice. Objective measures of offender classification provide a mechanism to challenge the conservative inclinations⁽¹²⁾ often pervading classification systems that rely too heavily on personal or subjective judgments and fall prey to many of the cognitive limitations associated with clinical judgment.⁽¹³⁾

Expeditiously placing as many offenders as possible at the lowest possible security level at the start of their sentence can minimize the effect of and contribute to the cost-effectiveness of managing prisons. Objective measures represent the corporate norms and provide the corporate authority for security classification decisions. This recognizes that the basic responsibility for classification protocols rest with the agency and, with a competent application of the protocols by the case management officer, the security classification exercise is less inclined to be influenced by conservative inclinations that often affect classification decisions. The objective measures provide a powerful tool to control the flow of offenders across security levels and could be an important impetus to the reintegration strategy that remains to be fully exploited.

A final word of caution is warranted, lest the impression is left that objective or actuarial measures are being advocated to the exclusion of what is popularly referred to as "professional judgment." On the contrary, objective measures are intended to inform professional judgment, not replace or overwhelm it. All too often, however, professional judgment is confused with what is more properly defined as "personal or subjective judgment," which is often influenced by personal taste, experience and prejudice. Subjective judgements are often antagonistic toward objective or actuarial approaches and antithetical to the notion of professional judgment. Judgment is professional to the extent that it embraces a theoretical framework that helps explain the behaviour in question; it is supported by a body of knowledge derived from empirical experience; it is represented by publicly explicit criteria and protocols that encourage fair and consistent decision making; and it has the public endorsement of and is enforced by its practitioners. Professional judgment requires a balance between unfettered subjectivity and blind acquiescence to actuarial tables. Within this context, objective measures can contribute both to the integrity of security classification practice and the success of the reintegration strategy.

---

1. P.O. Box 1174, Kingston, Ontario K7L 4Y8.

2. First Report on Progress for Federal/Provincial/Territorial Ministers for Justice (Fredericton, NB: Corrections Population Growth, 1997).

3. D. Hendrick, "Canadian Crime Statistics, 1995," Juristat, 16, 11 (1996).

4. Statistics Canada, Adult Correctional Services in Canada 1995-96 (Ottawa: Canadian Centre for Justice Statistics, 1997).

5. Correctional Service of Canada, The Task Force Report on Community and Institutional Programs (Ottawa: Correctional Service of Canada, 1988).

6. Correctional Service of Canada, "Contributing to Reintegration Through Appropriate Case Management Decisions and Improved Efficiency: A Discussion Paper" (Ottawa: Correctional Service of Canada, 1997).

7. R. A. Buchanan, K. L. Whitlow and J. Austin, "National evaluation of objective classification systems: The current state of the art," Crime and Delinquency, 23 (July, 1986): 272-290.

8. J. L. Proctor, "Evaluating a modified version of the Federal Prison System's inmate classification model: An assessment of objectivity and predictive validity," Crime and Delinquency, 21 (June, 1994): 256-272. See also G. D. Walters, Foundations of Criminal Science, 2 (New York: Praeger, 1992). And see J. Bonta and L. L. Motiuk, "Inmate classification," Journal of Criminal Justice, 20 (1992): 343-353. And see C. S. Baird and J. Austin, Current State of the Art in Prison Classification Models: A Literature Review for the California Department of Corrections (San Francisco, CA: NCCD, 1985). And see J. Austin, "Assessing the new generation of prison classification models," Crime and Delinquency, 29 (October 1983): 561-577.

9. F. P. Luciani, L. L. Motiuk and M. Nafekh, An Operational Review of the Custody Rating Scale: Reliability, Validity and Practical Utility, Report 47 (Ottawa: Correctional Service of Canada, 1996).

10. F. P. Luciani, L. L. Motiuk and N. Mainwaring, Field Tests of the Custody Rating Scale (Ottawa: Correctional Service of Canada, in press). See also F. P. Luciani and L. L. Motiuk. Security Classification of Offenders: Development of Reclassification Protocols. (Ottawa: Correctional Service of Canada, in press).

11. Austin, "Assessing the new generation of prison classification models."

12. Solicitor General of Canada, Development of a Security Classification Model for Canadian Federal Offenders (Ottawa: Correctional Service of Canada, 1987).

13. W. M. Grove and P. E. Meehl, "Comparative efficiency of informal (subjective, impressionistic) and formal (mechanical, algorithmic) prediction procedures: The clinical-statistical controversy," Psychology, Public Policy and Law, 2, 2 (1996): 293-323. See also R. T. Palmer, Predicting Release Outcome Using Current Psychological and Behavioural Variables, a Longitudinal Criterion Measure and Event History Analysis. Unpublished Doctoral Dissertation (Kingston, ON: Queen's University, 1997). And see V. L. Quinsey, G. T. Harris and M. E. Rice, Violent Offenders: Appraising and Managing Risk (In preparation, 1997).